Deliverable D3.7a: Road Safety Performance Indicators Country Comparisons Please refer to this report as follows: Vis, M.A. and Van Gent, A.L. (Eds.) (2007) Road Safety Performance Indicators: Country Comparisons. Deliverable D3.7a of the EU FP6 project SafetyNet. Contract No: TREN-04-FP6TR-S12.395465/506723 Acronym: SafetyNet Title: Building the European Road Safety Observatory Integrated Project, Thematic Priority 6.2 Sustainable Surface Transport Project Co-ordinator: Professor Pete Thomas Vehicle Safety Research Centre Ergonomics and Safety Research Institute Loughborough University Holywell Building Holywell Way Loughborough LE11 3UZ Organisation name of lead contractor for this deliverable: SWOV Editors: Martijn A. Vis and Alex van Gent (SWOV) Report Author(s): Kerstin Auerbach-Hafen (BASt); François Riguelle (IBSR); Vojtech Eksler (CDV); Mouloud Haddak (INRETS); Péter Holló (KTI); Elisabete Arsenio, João Cardoso, Sandra Vieira Gomes (LNEC); Eleonora Papadimitriou (NTUA); Maarten Amelink, Charles Goldenbeld, René Mathijssen, Robert Louwerse, Peter Morsink, Chris Schoon, Alex van Gent, Sjoerd Houwing, Martijn Vis (SWOV); Victoria Gitelman, Shalom Hakkert (TECHNION); Terje Assum (TØI); Marianne Page, Lucy Rackliff (VSRC) Due Date of Deliverable: 31/12/2006 Submission Date: 31/12/2006 Project Start Date: 1st May 2004 Duration: 4 years Project co-funded by the European Commission within the Sixth Framework Programme (2002-2006) Dissemination Level PU Public Project co-financed by the European Commission, Directorate-General Transport and Energy
Executive Summary This report compares the safety performance of 27 European countries the25 EU member states, Norway and Switzerland. The comparison is done for seven road safety related areas: alcohol and drugs, speeds, protective systems, daytime running lights, vehicles (passive safety), roads, and trauma management, on basis of the theory presented in Hakkert, Gitelman and Vis 1 (2007), using the data obtained from the collaborating countries (see Vis and Van Gent 2 (2007). When indicator values are available but not comparable due to e.g. lack of data quality, this is explained. In general, comparing the countries' performances is difficult. The main reasons are the lack of data, suspicious quality of the data, or the incomparability of the (seemingly similar) data due to different circumstances of measurement. As an example of the latter, one might think of speed measurements for different road types in different countries, or on similar road types with completely different characteristics. In a number of cases, the choice for a specific performance indicator depends on the availability of data. This has, for example, been the case for the indicator for alcohol usage; while the optimal indicator would concern the usage rate of alcohol in the general driver population, the unavailability of data in a number of the (larger) country, has led to a more indirect indictor. Details about the development of the safety performance indicators can be found in Hakkert, Gitelman and Vis (2007). In spite of all considerations and limitations, we are able to present a great number of comparisons in this report, or to present the figures that can form the basis for future comparisons. Reliable comparisons are made for the areas daytime running lights, protective systems, vehicles (passive safety), and trauma management. Only limited comparisons are made for the areas speeds and roads. Due to great differences in data quality between the different countries, comparisons in the area alcohol and drugs is not possible. The results for that area are presented for information only and will form the basis for future study. 1 Hakkert, A.S., Gitelman, V., and Vis, M.A. (Eds.) (2007) Safety Performance Indicators: Theory. Deliverable D3.6 of the EU FP6 project SafetyNet. 2 Vis, M.A., and Van Gent, A.L. (Eds.) (2007) Safety Performance Indicators: Country Profiles. Deliverable D3.7b of the EU FP6 project SafetyNet. sn_wp3_d3p7a_spi_country_comparisons Page 2
Contents 1 Introduction...5 2 Development of Safety Performance Indicators in SafetyNet...6 2.1 Definition of safety performance indicators (SPIs)...6 2.2 The role of SPIs in safety management...6 2.3 Selected areas for SPI development...7 2.4 Optimal indicators where possible, indirect indicators where necessary...7 2.5 General overview of SPIs developed and used for comparisons...8 3 Data availability overview...10 3.1 Data usability and country profile presence...10 3.2 Data quality...11 4 Alcohol and drugs...12 4.1 Indicators used...12 4.2 Are comparisons possible?...12 4.3 Country comparisons...13 5 Speed...16 5.1 Indicators used...16 5.2 Are comparisons possible?...16 5.3 Country comparisons for motorways...17 6 Protective systems...21 6.1 Indicators used...21 6.2 Are comparisons possible?...21 6.3 Country comparisons...21 6.4 Considerations...25 7 Daytime running lights...26 7.1 Indicators used...26 7.2 Are comparisons possible?...26 7.3 Country comparisons...26 8 Vehicles (passive safety)...29 8.1 Indicators used...29 8.2 Are comparisons possible?...29 8.3 Country comparisons...29 9 Roads...38 9.1 Indicators used...38 sn_wp3_d3p7a_spi_country_comparisons Page 3
9.2 Are comparisons possible?...38 9.3 Reliability and representativeness...39 9.4 Country comparisons...39 10 Trauma management...47 10.1 Indicators used...47 10.2 Are comparisons possible?...47 10.3 Country comparisons...47 10.4 Considerations...51 11 Conclusions...53 References...54 Appendix A Estimation of overall seat belt wearing rates for indicators SPI-A and SPI-B...55 Appendix B Trauma management basic data...56 sn_wp3_d3p7a_spi_country_comparisons Page 4
1 Introduction This document presents the comparison of the road safety performance of 27 European countries the 25 EU member states, Norway and Switzerland. The performance of these countries is based on their score on so-called safety performance indicators in seven roadsafety related areas: alcohol and drugs, speed, protective systems, daytime running lights, vehicles (passive safety), roads, and trauma management. We developed road safety performance indicators for each of these areas in Hakkert, Gitelman and Vis (2007). Safety performance indicators are seen as any measurement that is causally related to crashes or injuries and is used in addition to the figures of accidents or injuries, in order to indicate safety performance or understand the process that leads to accidents (ETSC, 2001). They also provide the link between the casualties from road accidents and the measures to reduce them (ETSC, 2006). Safety performance indicators help illustrate how well road safety programs are doing in meeting their objectives or achieving the desired outcomes. They are a means of monitoring, assessing and evaluating the processes and operations of road safety systems concerning their potential to solve the problems they are up against. They use qualitative and quantitative information to help to determine a program's success in achieving its objectives. They could be used to track progress and could provide a basis to evaluate and improve performance. SafetyNet s Road Safety Performance project team has worked closely together with national representatives of the 27 countries to obtain as much of the data relevant for calculating the indicator values. A complete overview of all underlying data obtained for the 27 countries can be found in Vis and Van Gent (2007).The current report presents the indicator values as far as they were found suitable for comparison with other countries indicator values. In many cases, we found that essential data were missing or that the quality of the data was too poor to use for country comparisons. For example, this was the case for the areas related to alcohol and drugs use and to roads. Yet, even in these cases, we have often presented the indicator values, but explicitly stated the extent to which we found the comparisons valid. Chapter 2 first gives a brief instruction into the background of safety performance indicators, explaining their role in road safety management and the way to develop appropriate and feasible indicators. Next, it presents an overview of the indicators used for the country comparisons in this report. The next chapter (chapter 3) gives an overview of the data available to the SafetyNet Road Safety Performance Indicators team. The overview shows, per indicator area and per country, whether any data were obtained from the national representative (or from another source) and whether the data obtained was suitable for calculating the indicator values. Chapters 4 to 10 present the results for each indicator area consecutively. For each area, the developed safety performance indicator is briefly presented, and it is explained to what extent the available data allows a comparison of the countries performances. After this, the country comparisons are presented in terms of graphs and tables. For some indicator areas, some of the underlying data and methods are detailed in the appendices to this report. Finally, in chapter 11 we present our overall conclusions. Acknowledgement The authors would like to thank the National Experts of the 27 cooperating countries (25 member states, Norway and Switzerland) for providing the data and for giving feedback on concept versions of this report. sn_wp3_d3p7a_spi_country_comparisons Page 5
2 Development of Safety Performance Indicators in SafetyNet VIS A, M.A. A SWOV This chapter gives a brief overview of the theory behind road safety performance indicators (SPIs) and the way they were developed for the different road safety related areas, alcohol and drugs, speed, protective systems, daytime running lights, vehicles (passive safety), roads, and trauma management. The full theory behind the indicators developed can be found in Hakkert, Gitelman and Vis (2007). 2.1 Definition of safety performance indicators (SPIs) The following definition of safety performance indicators can be given: Safety performance indicators are the measures (indicators), reflecting those operational conditions of the road traffic system, which influence the system s safety performance. The purpose of safety performance indicators is to reflect the current safety conditions of a road traffic system (i.e. they are considered not necessarily in the context of a specific safety measure, but in the context of specific safety problems or safety gaps); to measure the influence of various safety interventions, but not the stage or level of application of particular measures, to compare between different road traffic systems (e.g. countries, regions, etc). 2.2 The role of SPIs in safety management Safety performance indicators are seen as any measurement that is causally related to crashes or injuries and is used in addition to the figures of accidents or injuries, in order to indicate safety performance or understand the process that leads to accidents (ETSC, 2001). They also provide the link between the casualties from road accidents and the measures to reduce them (ETSC, 2006a). Safety performance indicators can give a more complete picture of the level of road safety and can point to the emergence of developing problems at an early stage, before these problems show up in the form of accidents (ETSC, 2001; Luukkanen, 2003). Because of this, safety performance indicators help illustrate how well road safety programs are doing in meeting their objectives or achieving the desired outcomes. They are a means of monitoring, assessing and evaluating the processes and operations of road safety systems concerning their potential to solve the problems they are up against. They use qualitative and quantitative information to help to determine a program's success in achieving its objectives. They could be used to track progress and could provide a basis to evaluate and improve performance. In order to properly perform their function, SPIs need to be relevant to the program s desired outcomes and objectives, and to be quantifiable, verifiable and unbiased. sn_wp3_d3p7a_spi_country_comparisons Page 6
2.3 Selected areas for SPI development Following the recommendations of the ETSC report "Transport Safety Performance Indicators" (2001), seven problem areas were selected for the SPIs' development in SafetyNet. They are: Alcohol and drug-use Speeds Protection systems Daytime running lights (DRL) Vehicles Roads Trauma management Note that these seven domains are related to different levels of the road safety system. While alcohol and speeds address "road safety problems" (or unsafe system conditions), protection systems and DRL reflect countermeasures which are intended to prevent accidents ("DRL") or to lower accident consequences ("protective systems"). The domains roads and vehicles are related to a wide area of road safety interventions, whereas alcohol or speeds are related to the area of human behaviour as cause of accidents. The domain "rescue services" (trauma management) presents an additional category of road safety issues. 2.4 Optimal indicators where possible, indirect indicators where necessary Under normal circumstances the optimal indicator for an issue would be a direct indicator. Often this is not realizable, e.g. due to a lack of appropriate data. In that case indirect variables which describe the problem can be used as indirect indicators. If this is also not possible, the problem can be divided into several sub-problems and the indicator can be established for each of those. In this case the initial problem is not completely covered any more. Constructing composite indicators is possible, but difficult, because any weighting process is value-laden and perhaps no longer neutral. General methodology on constructing composite indicators is described in a handbook published by OECD (2005). In a number of cases, the SafetyNet road safety performance indicator team found that the available data do not allow for the optimal performance indicator to be used, so that a more indirect indicator had to be proposed and used. This was, for example, the case for the alcohol and drugs issue, where the optimal indicator would be related to the use of these substances in the general traffic population. Some of the larger countries in the EU, however, have laws prohibiting the appropriate measurement of the necessary data. This makes the optimal performance indicator undesirable. sn_wp3_d3p7a_spi_country_comparisons Page 7
2.5 General overview of SPIs developed and used for comparisons The following table gives an overview of the safety performance indicators developed by the SafetyNet team, and which were used for the country comparisons in this report. Indicator area Alcohol and drugs Developed indicator SPI-alcohol The percentage of fatalities resulting from accidents involving at least one driver impaired by alcohol SPI-drug The percentage of fatalities resulting from accidents involving at least one driver impaired by drugs other than alcohol Speeds The average speed either during daytime or during the night The percentage of speed limit offenders. Protective systems Daytime running lights Vehicles (passive safety) Roads Daytime wearing rates of seat belts A Front seats passenger cars + vans under 3.5 tons B Rear seats passenger cars + vans under 3.5 tons C Children under 12 years old - restraint systems use in passenger cars D Front seats heavy good vehicles (HGV) + coaches above 3.5 tons E Passenger seats - coaches Daytime wearing rates of safety helmets F Cyclists G Moped riders H Motorcyclists The total usage rate of daytime running lights The usage rate of daytime running lights per road type (4 types) The usage rate of daytime running lights per vehicle type (4 types) The crashworthiness and vehicle age of the passenger car fleet The vehicle fleet composition Network design Intersection types Intersection density Road design EuroRAP Road Protection Scores (RPS) Share of roads with a wide median or median barrier Share of roads with a wide obstacle-free zone or roadside barrier sn_wp3_d3p7a_spi_country_comparisons Page 8
Indicator area Trauma management Developed indicator (continued) Availability of Emergency Medical Services (EMS) stations The number of EMS stations per 10,000 citizens Availability and composition of EMS medical staff percentage of physicians and paramedics out of the total number of EMS staff The number of EMS staff per 10,000 citizens Availability and composition of EMS transportation units Percentage of Basic Life Support Units (BLSU), Mobile Intensive Care Units (MICU) and helicopters/planes out of the total number of EMS transportation units The number of EMS transportation units per 10,000 citizens The number of EMS transportation units per 100 km of total road length Characteristics of the EMS response time The demand for EMS response time (min) Percentage of EMS responses meeting the demand average response time of EMS (min) Availability of trauma beds in permanent medical facilities Percentage of beds in trauma centres and trauma departments of hospitals out of the total trauma care beds The total number of trauma care beds per 10,000 citizens Furthermore, a combined indicator was developed to measure a country's overall performance for trauma management. Table 2.1 Overview of the developed safety performance indicators per indicator area. sn_wp3_d3p7a_spi_country_comparisons Page 9
3 Data availability overview VAN GENT A, A. A SWOV In this chapter an overview is presented on the availability to the SafetyNet safety performance indicators team of data for each country for each of the seven indicator areas studied. In general, most countries provided a great deal of the data needed to calculate the safety performance indicators. For most indicator areas it is possible to make comparisons between several countries, based on the data provided. The first section in this chapter shows whether the available data are sufficient to calculate the safety performance indicators ('usability'). In the second section the quality of the data is briefly discussed. 3.1 Data usability and country profile presence On average, usable data for the calculation of safety performance indicators is available for two thirds of the countries. Most of these data were obtained via the National Experts, who filled in the questionnaires. For some indicator areas, the SafetyNet team has gathered missing or additional data via other channels, like the SUNflower+6 and the SARTRE projects. Table 3.1 gives an overview of the data availability and usability. For each country and for each indicator area the table indicates whether or not data are available and whether these data are suitable to calculate SPIs. Country Name Country code Alcohol and drugs Speed Protective systems DRL Vehicle Roads Trauma managment Belgium BE + + + - + + + Czech Republic CZ + + + + + + + Denmark DK + + + - + + + Germany DE + - + - + ~ + Estonia EE + + + + + + + Greece EL + - - - + + + Spain ES + ~ + - + + - France FR + + + + - - - Ireland IE - + + - - - - Italy IT + - + - ~ - - Cyprus CY + ~ - - + + + Latvia LV + ~ + - + ~ + Lithuania LT + - - - - - - Luxembourg LU - - + - - - - Hungary HU + + + + + + + Malta MT - - + - + ~ + The Netherlands NL + + + + + + + Austria AT + + + + + ~ + Poland PL + + + - ~ ~ - Portugal PT - + + - ~ + ~ Slovenia SI - - ~ - - - - Slovakia SK + - - - - - + Finland FI + + + + - - - Sweden SE + ~ + - + + + United Kingdom UK + + + - + - + Norway NO + + + - + + + Switzerland CH + + + + ~ - - + [green] = data available, and can be used for the calculation of performance indicators ~ [yellow] = data available, but not suitable for the calculation of performance indicators - [red] = no data available sn_wp3_d3p7a_spi_country_comparisons Page 10
Table 3.1. Overview data availability and presence of country profiles. Fortunately, each country did provide data for at least one of the areas. From the table it can be seen that the Czech Republic, Estonia, Hungary and The Netherlands have delivered data that is suitable to calculate safety performance indicators for all indicator areas. Many other countries could provide data for almost all indicator areas. Unfortunately there is also a small group of countries that could provide only a very small part of the data. Ireland, Italy, Lithuania, Luxembourg, Slovakia and Slovenia provided data for only one or two indicator areas. The great part of red cells in the column of DRL are notable. This is partly due to the fact that in a part of these countries (Denmark, Latvia, Norway, Sweden) the DRL usage rate is assumed to be already so high, that these countries do not find it interesting to determine the exact DRL usage rate. Furthermore there are countries in which the usage of DRL is not an issue, through which data on DRL usage rates are not collected. Taken into account that Denmark, Latvia, Norway, Sweden have such a high DRL usage rate that it is not interesting to measure the exact usage rate, these countries have provided data for all SPI areas. Most of these data are suitable for the calculation of SPIs. 3.2 Data quality A general issue is that the quality of the provided data is often unknown. It is clear that this affects the validity of the country comparisons. The received data were checked for basic omissions, like missing values. Nevertheless, a more thorough checking of the data quality will have to be done in the future. sn_wp3_d3p7a_spi_country_comparisons Page 11
4 Alcohol and drugs ASSUM A, T., HOLLÓ B, P., HOUWING C, S., MATHIJSSEN C, R. A TØI, B KTI, C SWOV 4.1 Indicators used The safety performance indicators used for alcohol and drugs, respectively, are: SPI-alcohol: The percentage of fatalities resulting from accidents involving at least one driver impaired by alcohol SPI-drugs: The percentage of fatalities resulting from accidents involving at least one driver impaired by drugs other than alcohol. 4.2 Are comparisons possible? To make realistic comparisons between countries, the statistics must be defined and collected in the same way in the countries to be compared. In the case of the SPIs above there are several differences between the countries which have provided data. Most countries provide data for drivers above the legal alcohol limit. As seen from the tables below this limit varies from 0.0 to 0.9 g/l blood alcohol concentration (BAC). The difference in legal limit may have two opposite effects. On the one hand the higher the limit, the lower the percentage of drivers who should be above this limit. On the other hand, if low legal limits have deterrent effects, there may be relatively less drivers above the legal limit in countries with low legal limits. Another important issue is the percentage of drivers involved in fatal accidents who are actually tested for alcohol and/or drugs. In France in 2005 the BAC level is known for the drivers in 4287 fatal accidents, whereas the total number of fatal accidents is 4857. In the example provided by the UK, however, about 41% of the fatal crashes have unknown driver BAC. If the police ask for blood samples only when there is a suspicion of a driver influenced by alcohol, the percentage of drivers above the legal limit will be higher among those tested than among those not tested. The question may then be asked whether the fatal accidents with alcohol-positive drivers should be related to the number of fatal accidents with drivers tested or to the total number of fatal accidents. Some countries include all fatal accidents where drivers under the influence have been involved, whereas others include only fatal accidents caused by drivers under the influence. The concept of cause is difficult in road accidents. Consequently, including only accidents caused by drivers under the influence may reduce the value of the indicator Especially for small countries the number of fatalities is small and is subject to random variation. To reduce the effects of random variation the safety performance indicators should preferably be computed based on data for several years, rather than for one year. Germany and Lithuania include fatalities from accidents involving drivers as well as bicycle riders and pedestrians under the influence of alcohol. Thus the values of the indicator for these two countries are likely to be higher than they would have been if only drivers under the influence were included. When comparing countries, the data should preferably be from the same year. However, this has not been possible in the case of fatal accidents with drivers under the influence of alcohol and drugs. Sixteen countries have been able to produce data for 2005, whereas the remaining six countries for which data were available have data from 2001 to 2004. On the one hand the prevalence of impaired drivers in a national population of drivers could be expected to be rather stable from one year to the next, and thus the number of fatal sn_wp3_d3p7a_spi_country_comparisons Page 12
accidents involving impaired drivers could also be expected to be rather stable over a few years, except for the random variation mentioned above. On the other hand one of the main objects of safety performance indicators is to show improvement or deterioration in safety over time, especially if some countries are making efforts to reduce unsafe conditions such as drinking and driving. Only six countries have provided data for drugs other than alcohol, and even some of these countries state that these figures are unreliable because very few drivers are tested for drugs. Data for drugs or psychoactive substances other than alcohol is a much more complex issue than alcohol. The number of drugs is large, and some drugs may be used in some countries but not in others. Drugs vary from medical drugs in prescribed doses, to medical drugs in abuse doses and to illicit drugs in varying doses. Drugs may be combined with each other or with alcohol. In some countries the police will only ask for drug testing if the alcohol breath test is negative, but still reason to suspect the presence of psychoactive substances. Only a few countries have legal limits for some drugs. Most countries having provided data for drugs, describe neither which drugs nor the limits that were used when considering a person under the influence. The above-mentioned factors causing differences may not be the only ones. Anyhow, they are too many to make simple transformations or corrections to improve comparability. Thus, the results shown below should be considered more as an example of possible results rather than as results showing actual and true differences between the countries. The differences between the countries are likely to reflect differences in data collection procedures and other methodological aspects more than differences in importance of alcohol and drugs as risk factors in road traffic in these countries. Producing reliable and valid, and thus comparable safety performance indicators for alcohol and drugs for the 27 countries is likely to require considerable efforts in harmonizing definitions, data collection and data analysis methods. The most important aspect is likely to be the number of drivers involved in fatal accidents, who are actually tested for alcohol and/or drugs. Each country should report the number of tested and untested drivers involved in fatal accidents in addition to the total number of fatalities and the number of fatalities resulting from accidents with at least one driver impaired by alcohol or drugs. 4.3 Country comparisons This section presents country comparisons for the alcohol and the drugs performance indicator, respectively. Results are only shown for those countries that have provided usable data. The following countries have supplied no data that could be used for computing the safety performance indicators for alcohol and drugs: Ireland, Luxembourg, Malta, Portugal, and Slovenia. 4.3.1 Alcohol Twenty-three of the 27 countries provided data that could be used to calculate the safety performance indicator for alcohol. Since the blood alcohol concentration (BAC) limit is an essential variable for the interpretation of the alcohol performance indicator, the countries are grouped according to BAC limit. Note that the measurement year can differ by country as discussed above. Country Year SPI-alcohol (%) BAC limit (g/l) Czech rep. 2004 4.8 0.0 Hungary 2005 8.7 0.0 Slovakia 2005 12.9 0.0 Table 4.1 Comparison of the alcohol safety performance indicator for countries with a BAC limit of 0.0 g/l. sn_wp3_d3p7a_spi_country_comparisons Page 13
The Czech Republic has a low performance indicator value for alcohol, in spite of its legal BAC limit of 0.0 g/l, and in spite of the fact that the Czech Republic has the second highest per capita alcohol consumption in Europe (WHO 2004). The question may be asked whether this low indicator reflects the real situation or if there is a methodological explanation to it. The Czech national experts have confirmed that all motor vehicle drivers involved in fatal accidents are tested for alcohol. Thus, unless there are other methodological explanations, it seems that the Czech Republic has achieved a situation of practical segregation of drinking and driving for which most other countries aim. Country Year SPI-alcohol (%) BAC limit (g/l) Comment Estonia 2005 23.5 0.2 Poland 2005 9.8 0.2 Sweden 2005 25.0 0.2 Estimated on the basis of autopsies of killed drivers. Norway 2001-2002 (22.2) 0.2 Killed drivers impaired by alcohol in % of all killed drivers rather than fatalities in accidents involving drivers impaired by alcohol. As alcohol-impaired drivers are over-represented in single-vehicle accidents, the figure for Norway is likely to be higher than the indicator for alcohol. Table 4.2 Comparison of the alcohol safety performance indicator for countries with a BAC limit of 0.2 g/l. Country Year SPI-alcohol (%) BAC limit (g/l) Comment Belgium 2002 8.2 0.5 Only an estimated 20% of drivers involved in fatal accidents are tested. If this estimation is taken into account, the indicator will be 40.7%. Denmark 2005 16.0 0.5 Greece 2004 9.4 0.5 Spain 2005 (29.5) 0.5 Killed drivers impaired by alcohol in % of all killed drivers rather than fatalities in accidents involving drivers impaired by alcohol. As alcohol-impaired drivers are over-represented in single-vehicle accidents, the figure for Spain is likely to be higher than the indicator for alcohol. France 2005 28.8 0.5 Calculated as % of fatal accidents with tested drivers. Likely to give higher value than if calculated as % of all fatal accidents. Hungary 2005 8.4 0.5 Legal limit 0.0, but data also provided for BAC>0.5. Latvia 2005 21.7 0.5 The 2005 8.3 0.5 Netherlands Austria 2005 5.9 0.5 Portugal 2005 (27.8) 0.5 Killed drivers impaired by alcohol in % of all killed drivers rather than fatalities in accidents involving drivers impaired by alcohol. As alcohol-impaired drivers are over-represented in single-vehicle accidents, the figure for Portugal is likely to be higher than the indicator for alcohol. Finland 2005 23.4 0.5 Switzerland 2005 19.3 0.5 Table 4.3 Comparison of the alcohol safety performance indicator for countries with a BAC limit of 0.5 g/l. Also for Austria the indicator value is so low that the question may be asked whether it conveys the real situation or whether it is due to methodological factors. sn_wp3_d3p7a_spi_country_comparisons Page 14
Four of the remaining countries all use a different BAC limit, and one country, Italy, has not provided information about its legal limit. Because of this difference, their calculated performance indicators are considered incomparable. Calculated values for the performance indicator for those countries are presented in the following table. Country Year SPI-alcohol (%) BAC limit (g/l) Comment Germany 2004 12.1 0.3 0.3 g/l is BAC limit of accident involved drivers Italy 2004 (72.2) 0.5 Extreme value. Request for confirmation submitted, but no reply received. Cyprus 2005 22.5 0.9 Limit changed to 0.5 in 2006 Lithuania 2005 14.8 0.4 UK 2004 17.0 0.8 Estimated by Department of Transport, UK Table 4.4 Calculated of the alcohol safety performance indicator for countries with BAC limits other than 0.0, 0.2 or 0.5 g/l. The value for the indicator for alcohol varies from 4.8% in the Czech republic to 28.8% in France, disregarding the figures for Spain and Italy, which in the case of Spain is a slightly different statistic and in the case of Italy is likely to be an error. However, the French result is likely to be an overestimation as it is computed on the basis of the fatalities for which the BAC level of the drivers was known. For Belgium the value of the indicator is 8.2%, but it is estimated that only some 20% of drivers involved in fatal accidents are tested for alcohol. If this fact is taken into consideration, the indicator value for Belgium may be higher. This example shows that extreme care should be taken in comparing the alcohol performance indicator values for the European countries at this stage. 4.3.2 Drugs Only few countries could provide data that could be used to calculate the value of the performance indicator for drugs. The following table provides an overview of this indicator for those countries. Country Year SPI-drug (%) Belgium 2002 0.9 Czech rep. 2004 0.1 Cyprus 2005 2.9 Finland 2005 1.8 Norway* 2001-2002 (30.1) Switzerland 2005 7.6 Table 4.5 Comparison of the drugs safety performance indicator. (*The figure for Norway is the number of killed drivers impaired by drugs as percentage of all killed drivers, which is likely to yield an overestimation of the indicator value. See also the comment to the Norway data in Table 4.2.) As described in section 4.2 only one of the countries providing data for drugs (Switzerland) describes which drugs or which limits used, and only for illegal drugs. Consequently, the figures in Table 4.5 should be considered as an example of the drug safety performance indicator rather than comparable figures. sn_wp3_d3p7a_spi_country_comparisons Page 15
5 Speed RIGUELLE A, F. A IBSR 5.1 Indicators used The safety performance indicators used for speed are: The average speed either during daytime or during the night; The percentage of speed limit offenders. 5.2 Are comparisons possible? The possibility of international comparisons for speeding performance is very limited at the moment. The problem is not the availability of speed data throughout Europe; most countries make large-scale speed surveys and even compute the proposed safety performance indicators. The average speed and the percentage of offenders are the more frequently reported indicators. The huge variability in the way countries conduct their survey, however, prohibits wide-scale comparison. Main issues are: Representativeness of measuring locations Only 8 out of the 18 European countries which have data and for which we have information use a sampling procedure to select their measuring locations. The others prefer to choose them on high traffic or high accident rate axes only. They are sometimes even not interested in a national estimate and concentrate on individual road analysis. Furthermore, in some countries (e.g., Germany, The Netherlands), it is not a national organism which is responsible for speed monitoring. That leads to different types of speed surveys in different parts of the country, producing data that are impossible to aggregate at the national level. Unfortunately, we do not have estimates of the influence of the choice of the measuring location on the value of the safety performance indicators. Traffic conditions The traffic conditions under which the measurements are considered as valid also vary across countries. Ireland and Austria select perfectly free-flowing vehicles only. The United Kingdom only leaves out obvious congestion periods. Others countries lay in between (Belgium and the Czech Republic) or do not give information. Since traffic conditions have a significant impact on the speeds at which drivers operate their vehicles, one should only compare speed measurements that were carried out in similar non-congested traffic conditions. This is not strictly possible at the moment due to the differences in methodologies between countries. Comparability of roads Road classifications and speed limits vary between countries. The information on this is generally well-reported along with survey results but issues of comparability are still raised. It is not possible to define a simple transformation rule that would allow comparing similar roads with different speed limits in a perfect way. It is also impossible to find one corresponding road in each country for each SafetyNet road category. Nevertheless, three road types can be found almost anywhere: motorways (AAA), single carriageways A-level road (A) and urban single carriageway distributor roads (D). Even roads of the same type and of the same speed limit are designed differently across Europe. The influence of that can be minimised if strict criteria are applied to select sn_wp3_d3p7a_spi_country_comparisons Page 16
measuring locations. There is a broad acceptance of the idea that speed measurement should be done on straight roads and far from anything that may slow speed. However, the effort made to apply these criteria may vary, especially for urban roads where it is very hard to meet all the criteria. The information on the 'freedom' taken by surveyors in relation to their 'ideal' criteria is never available in the survey reports. Period of measurement The length of time of measurements varies from a few hours to a whole year depending on the country. When speed is measured for a few hours, this is mostly done during the day. Distinction between day and night measurements is usually available for longer studies but other time distinctions (weekday/weekend, time of the year) are more uncommon. Vehicle types Speed indicators are not published everywhere for the same types of vehicles. Due to the different shares in traffic of the different types of vehicles between countries, it is better to compare indicators for one vehicle type only (e.g. cars). Unfortunately, indicators aggregated over all vehicles types are sometimes the only available indicators. Accuracy of data There are many sources of uncertainties in speed data: accuracy of the device, representativeness of the sample of locations, size of the sample, handling of data, etc. It is thus virtually impossible to calculate the margin of error on the finally calculated safety performance indicators. In comparisons it will be impossible to determine with certitude whether any two values are significantly different or not. 5.3 Country comparisons for motorways Despite all these restrictions, we propose a comparison of speeds on motorways. On this type of road, the issue of road comparability is minimised (but speed limits still differ). Intuitively, it seems also easier for countries to produce a representative sample of their motorway network comparing to other road types. Figure 5.1 and Figure 5.2 illustrate the comparisons. Values for daytime only, night time only and whole days are reported in the same graphs. It must also be noted that no standardisation was carried out concerning different speed limits across countries. The different speeds limits are indicated by different colours (note the speed limit change in Ireland between 2003 and 2005. The comparison is limited to 2005 because annual indicators for 2006 data are not yet available except in Czech Republic. sn_wp3_d3p7a_spi_country_comparisons Page 17
Average Speed 130 125 FR - day Speed limit 130 km/h 120 km/h 112.6 km/h Speed (km/h) 120 115 110 FR - night AT CH UK PT - day PT - night CZ 105 IE day 100 1999 2000 2001 2002 2003 2004 2005 Year Figure 5.1 Comparison of average speeds of cars* on motorways between six countries (* all traffic together for Czech Republic and Switzerland). For most of the differences in methodologies between countries, we have a qualitative idea of the influence it should cause on measured speeds. For example, the fact that Ireland measures strictly free-flowing vehicles will cause the measured speeds to increase; motorways with higher speed limits will have higher speeds. We will not comment differences that are following our expectations because we cannot estimate whether the observed differences are only caused by differences in methodologies or not. On the other hand, if differences are contradictory compared to our expectations, it is more relevant because it means that additional factors, such as the behaviour of drivers related to speed, may differ between the countries. The figure shows that average speeds are relatively low in Ireland compared to the other countries. Until 2004, the Irish speed limit is comparable to that in the UK and higher than on French 110 km/h roads but Irish speeds are significantly lower. In 2005, despite the raise of the speed limit in Ireland to 120 km/h, the average speed remained lower than on UK motorways and is the same as on French 110 km/h motorways. The inverse results would have been expected, because the UK and France only leave out obvious congestion conditions from their data but do not select perfectly free-flowing vehicles, such as it is done in Ireland. Despite higher speed limits than in the UK, average speeds in Switzerland and Czech Republic are slightly lower in the last years. But it should be noted that all vehicles types are included in the indicators for these two countries, which likely have the consequence of lowering the average speed comparing to a car-only situation. sn_wp3_d3p7a_spi_country_comparisons Page 18
We may also look at the trends. First note that speed can vary much over the years. This may be due to actual changes in speed but also to variability generated by the methodologies. Whatever the cause, it implies that comparisons should not be based on one single year of measurement but better on a time series of data. It also shows the interest for countries to implement regular speed measurements instead of episodic surveys. We saw that speeds decrease on French motorways, contrarily to Austria, UK and Ireland. The strong emphasis made by French authorities on enforcement is surely the main cause of this observation. In summary, the only things we can mention with the comparisons are that the speeds on Irish motorways are low and that it is in France that the most progress is currently made. It would be very hazardous to try to derive more conclusions from the current data. In general, current speed data are more accurate to compare trends than absolute values because internal country methodologies usually remain consistent in time. On the other hand, nothing can be stated with enough certitude at the moment about absolute values due to the difference in the methodologies used across counties. On a side note, French and Portuguese data also allow studying the difference in speeds between daytime and night. Unexpectedly, speeds at night are lower than daytime speeds in Portugal and on French 130 km/h motorways. We may consider that the observed differences between day and night speeds in France are significant because they remain over time and at the scale of one country, the methodologies for day and night measures are the same. This finding reemphasises the idea that day and night speeds should be considered separately and should not be combined into one safety performance indicator. % of cars over the limit 60 50 UK FR day Speed limit 130 km/h 120 km/h % over limit 40 30 20 FR night CH AT 112.6 km/h 10 IE - day 0 1999 2000 2001 2002 2003 2004 2005 Year Figure 5.2 Comparison of the percentage of cars* over the limit on motorways in five counties (* all traffic together for Switzerland). sn_wp3_d3p7a_spi_country_comparisons Page 19
Concerning the percentage of offenders, we observe that its value is higher on roads with lower speed limits. Still, Ireland is an exception. The percentage tends to decrease on French motorways in the last three years. This is also the case in Ireland. sn_wp3_d3p7a_spi_country_comparisons Page 20
6 Protective systems EKSLER A, V. A CDV 6.1 Indicators used The safety performance indicators used for protective systems are: Daytime wearing rates of seat belts A Front seats passenger cars + vans under 3.5 tons B Rear seats passenger cars + vans under 3.5 tons C Children under 12 years old - restraint systems use in passenger cars D Front seats heavy good vehicles (HGV) + coaches above 3.5 tons E Passenger seats - coaches Daytime wearing rates of safety helmets F Cyclists G Moped riders H Motorcyclists 6.2 Are comparisons possible? A comparison can be performed on basis of the data available to the SafetyNet team. The use of at least one type of protective system in traffic has been assessed through independent roadside surveys in almost all countries, except Cyprus, Greece, Lithuania and Slovakia. In some particular cases, the values of indicators cannot be considered as valid and comparable as they do not fulfil defined conditions on accuracy. In these cases, the rough estimates of indicators are presented to give an impression of the magnitude of relevant road safety problem. All results presented concern the situation in 2005, unless stated otherwise. Data for 2006 will becomes available for many countries only in early 2007. The following criteria were considered regarding the validity and comparability of the indicators produced by member states: Origin - roadside observational survey (independent) Fitness to indicator/sample definition (road user definition) Representativeness (observations on all road types, several locations for each) Time coverage - daytime during week days, no public holidays period Appropriate aggregation (by exposure) + transformation rules applied 6.3 Country comparisons 6.3.1 SPI-A: daytime wearing rates of seat belts in front seats of passenger cars and vans under 3.5 tons The seat belt wearing rate in front seats (whether driver only, or also front passenger) is assessed in 20 of the 25 EU member states, in Norway and in Switzerland. The rates for France, Italy, Latvia, Malta, Poland and Portugal cannot be considered as valid and fully comparable, because they do not fulfil all above-mentioned criteria. (In particular, they are sn_wp3_d3p7a_spi_country_comparisons Page 21
usually not representative for the entire road network due to a limited number of observation sites on one/two road types only.) However, they still provide a rough estimate of indicator value. For many countries, only disaggregated values for driver and front seat passenger are available. Where an aggregated value was not available, we made use of a weighting coefficient of 0,35 for front seat passenger and 0,65 for the driver to get the value of the desired indicator (unless stated otherwise). (By the way, in countries with the most sophisticated survey design such as Germany, France, Sweden, or UK, the wearing rates for driver and for front seat passenger vary by max.1-2 %-points.) Similarly, if the rate was available for the driver only, it has been considered as corresponding to the indicator SPI-A. SPI value in % 100 90 80 70 60 50 40 30 20 10 0 SPI-A 96 97 96 90 85 86 71 72 74 74 77 80 83 86 88 92 90 90 82 78 71 67 BE CZ DK DE EE EL ES FR IE IT CY LV LT LU HU MT NL AT PL PT SI SK FI SE UK NO CH Figure 6.1 2005 Daytime seat belt wearing rate on front seats of passenger cars and vans under 3.5 tons (SPI-A). Remarks: LU: 2003; LV,MT: 2006; DK,DE,EE,IT,FR,PT,LU,CH: only driver wearing rates considered; FR: vans not included; IT,LV,MT,PL,PT does not fit to defined requirements. Only Germany, France and Malta register wearing rates above 95%, while the rates under 75% are registered in Belgium, Czech Republic, Estonia, Spain, Hungary and Poland. The rates in Slovakia, Lithuania and Greece, where the surveys have not been performed yet are presumably even lower, as foreshadowed by available data on the indirect indicator (rates by accident fatalities). 100 90 80 70 60 50 40 30 20 10 0 1985 1986 Evolution of SPI-A in time 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 ` 1999 2000 2001 2002 2003 2004 2005 2006 Figure 6.2 Time series of daytime seat belt wearing rate on front seats of passenger cars and vans under 3.5 tons (SPI-A). sn_wp3_d3p7a_spi_country_comparisons Page 22 BE DK EE FR IT LV HU NL PL FI UK CH CZ DE ES IE CY LU MT AT PT SE NO
From the time-series data, the evaluation of countries' seat belt wearing rates in time can be done, however an attention must be paid, since the national methodologies were often subject to changes over the years. The concept of conversion rates originally proposed by the NHTSA is used here. The conversion rate is the rate of decrease of non-use of protective systems from one year to other. The countries with already high wearing rates are not penalized by this evaluation. The user/nonuser categorization is a bit simplistic, since most of vehicle occupants are part-time users. However the use/non-use categorization is helpful for thinking about conversion rates. Here we can evaluate the improvement in seat belt wearing (SPI-A indicator) realized between 2000 and 2005. This was highest in The Netherlands (CR=55%), Hungary (CR=40%), Norway (CR=38%) and Czech Republic (CR=30%). That is, the countries converted 55%, 40%, 38%, and 34%, respectively, of its population that was not using belts in front seats in 2000 to using belts in 2005. We can further estimate the overall wearing rates in 25 EU member states using known and estimated indicator values and weighting them by the exposure of country population in road traffic (EUROSTAT). This is determined as 86% in 2005. (See Appendix A.) 6.3.2 SPI-B: daytime wearing rates of seat belts in rear seats of passenger cars and vans under 3.5 tons The seat belt wearing by passengers in rear seats of passenger cars and vans is assessed in 16 EU member states, in Norway and in Switzerland. Rates for the Czech Republic, Latvia and Malta cannot be considered as valid and fully comparable, but still provide a rough estimation of CRS use in road traffic. SPI value in % 100 90 80 70 60 50 40 30 20 10 0 SPI-B 89 84 85 78 73 70 63 64 60 51 52 53 46 45 41 34 30 28 BE CZ DK DE EE EL ES FR IE IT CY LV LT LU HU MT NL AT PL PT SI SK FI SE UK NO CH Figure 6.3 2005 Daytime seat belt wearing rate on rear seats of passenger cars and vans under 3.5 tons by persons above 12 years old (SPI-B). Remarks: LU: 2003, CZ, LV, MT: 2006; DK>16 years old, AT, IE>18 years old. For all countries, the rates are substantially lower in comparison with the wearing rates on front seats (SPI-A) and in general are higher in those countries with higher rates for front seats and lower in those countries with lower rates for front seats. More precisely, there is a large positive correlation between the values of indicators SPI-A and SPI-B. (For 16 data pairs, the Pearson's product-moment correlation coefficient is equal to 0.65, p- value=0.0066.) Improvement done between 2000 and 2005 was highest in Netherlands (CR=47%), Germany (CR=39%) and Switzerland (CR=31%). That is, the countries converted 47%, 39%, and 31%, respectively, of its population that was not wearing seat belts in rear seats in 2000 to using belts in 2005. We can further estimate the overall wearing rate in EU25 member states using weights for traffic performance (EUROSTAT). This is determined as 63% in 2005. (See Appendix A.) sn_wp3_d3p7a_spi_country_comparisons Page 23